always state encoding when opening files in text mode

[nominatim.git] / nominatim / tools / replication.py

diff --git a/nominatim/tools/replication.py b/nominatim/tools/replication.py
index 04f1c45b9cd728b7b0abe7cbb5ff892eb0fd7bef..53571706472036f80b546846f23b43fc905ec6c6 100644 (file)
--- a/nominatim/tools/replication.py
+++ b/nominatim/tools/replication.py
@@ -1,16 +1,29 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# This file is part of Nominatim. (https://nominatim.org)
+#
+# Copyright (C) 2022 by the Nominatim developer community.
+# For a full list of authors see the git log.
 """
 Functions for updating a database from a replication source.
 """
+from contextlib import contextmanager
 import datetime as dt
 from enum import Enum
 import logging
 import time
 
-from osmium.replication.server import ReplicationServer
-from osmium import WriteHandler
+from nominatim.db import status
+from nominatim.tools.exec_utils import run_osm2pgsql
+from nominatim.errors import UsageError
 
-from ..db import status
-from .exec_utils import run_osm2pgsql
+try:
+    from osmium.replication.server import ReplicationServer
+    from osmium import WriteHandler
+except ImportError as exc:
+    logging.getLogger().fatal("pyosmium not installed. Replication functions not available.\n"
+                              "To install pyosmium via pip: pip3 install osmium")
+    raise UsageError("replication tools not available") from exc
 
 LOG = logging.getLogger()
 
@@ -31,11 +44,11 @@ def init_replication(conn, base_url):
         LOG.fatal("Cannot reach the configured replication service '%s'.\n"
                   "Does the URL point to a directory containing OSM update data?",
                   base_url)
-        raise RuntimeError("Failed to reach replication service")
+        raise UsageError("Failed to reach replication service")
 
     status.set_status(conn, date=date, seq=seq)
 
-    LOG.warning("Updates intialised at sequence %s (%s)", seq, date)
+    LOG.warning("Updates initialised at sequence %s (%s)", seq, date)
 
 
 def check_for_updates(conn, base_url):
@@ -80,7 +93,7 @@ def update(conn, options):
     if startseq is None:
         LOG.error("Replication not set up. "
                   "Please run 'nominatim replication --init' first.")
-        raise RuntimeError("Replication not set up.")
+        raise UsageError("Replication not set up.")
 
     if not indexed and options['indexed_only']:
         LOG.info("Skipping update. There is data that needs indexing.")
@@ -97,22 +110,39 @@ def update(conn, options):
         options['import_file'].unlink()
 
     # Read updates into file.
-    repl = ReplicationServer(options['base_url'])
+    with _make_replication_server(options['base_url']) as repl:
+        outhandler = WriteHandler(str(options['import_file']))
+        endseq = repl.apply_diffs(outhandler, startseq + 1,
+                                  max_size=options['max_diff_size'] * 1024)
+        outhandler.close()
 
-    outhandler = WriteHandler(str(options['import_file']))
-    endseq = repl.apply_diffs(outhandler, startseq,
-                              max_size=options['max_diff_size'] * 1024)
-    outhandler.close()
+        if endseq is None:
+            return UpdateState.NO_CHANGES
 
-    if endseq is None:
-        return UpdateState.NO_CHANGES
+        # Consume updates with osm2pgsql.
+        options['append'] = True
+        options['disable_jit'] = conn.server_version_tuple() >= (11, 0)
+        run_osm2pgsql(options)
 
-    # Consume updates with osm2pgsql.
-    options['append'] = True
-    run_osm2pgsql(options)
-
-    # Write the current status to the file
-    endstate = repl.get_state_info(endseq)
-    status.set_status(conn, endstate.timestamp, seq=endseq, indexed=False)
+        # Write the current status to the file
+        endstate = repl.get_state_info(endseq)
+        status.set_status(conn, endstate.timestamp if endstate else None,
+                          seq=endseq, indexed=False)
 
     return UpdateState.UP_TO_DATE
+
+
+def _make_replication_server(url):
+    """ Returns a ReplicationServer in form of a context manager.
+
+        Creates a light wrapper around older versions of pyosmium that did
+        not support the context manager interface.
+    """
+    if hasattr(ReplicationServer, '__enter__'):
+        return ReplicationServer(url)
+
+    @contextmanager
+    def get_cm():
+        yield ReplicationServer(url)
+
+    return get_cm()